Data mining for decision support with uncertainty on the airplane

This study describes the formalization of the medical decision-making process under uncertainty underpinned by conditional preferences, the theory of evidence and the exploitation of high-utility patterns in data mining. To assist a decision maker, the medical process (clinical pathway) was implemented using a Conditional Preferences Base (CPB). Then for knowledge engineering, a Dempster-Shafer ontology integrating uncertainty underpinned by evidence theory was built. Beliefs from different sources are established with the use of data mining. The result is recorded in an In-flight Electronic Health Records (IEHR). The IEHR contains evidential items corresponding to the variables determining the management of medical incidents. Finally, to manage tolerance to uncertainty, a belief fusion algorithm was developed. There is an inherent risk in the practice of medicine that can affect the conditions of medical activities (diagnostic or therapeutic purposes). The management of uncertainty is also an integral part of decision-making processes in the medical field. Different models of medical decisions under uncertainty have been proposed. Much of the current literature on these models pays particular attention to health economics inspired by how to manage uncertainty in economic decisions. However, these models fail to consider the purely medical aspect of the decision that always remains poorly characterized. Besides, the models achieving interesting decision outcomes are those considering the patient's health variable and other variables such as the costs associated with the care services. These models are aimed at defining health policy (health economics) without a deep consideration for the uncertainty surrounding the medical practices and associated technologies. Our approach is to integrate the management of uncertainty into clinical reasoning models such as Clinical Pathway and to exploit the relationships between the determinants of incident management using data mining tools. To this end, how healthcare professionals see and conceive uncertainty has been investigated. This allowed for the identification of the characteristics determining people under uncertainty and to understand the different forms and representations of uncertainty. Furthermore, what an in-flight medical incident is and how its management is a decision under uncertainty issues was defined. This is the first phase of common data mining that will provide an evidential transaction basis. Subsequently an evidential and ontological reasoning to manage this uncertainty has been established in order to support decision making processes on the airplane.